2024 – Present

Research Scientist

SRON Space Research Organization Netherlands (The Netherlands)

Climate models, such as ECHAM6-HAM and ICON-HAM, are essential tools for understanding past and future climate change. However, significant uncertainties remain, particularly in the representation of aerosols and clouds. My work focuses on reducing these uncertainties by combining advanced climate modelling, perturbed parameter ensembles (PPEs), machine learning, and new satellite observations from missions including NASA/SRON's PACE and ESA's EarthCARE.

To achieve this, I develop perturbed parameter ensembles consisting of hundreds of climate model simulations with systematic variations in model parameters. Using these PPEs as training sets, I apply machine learning techniques to emulate millions of additional model configurations efficiently. This approach enables us to identify the sources of uncertainty within climate models, evaluate model outputs against satellite observations, and constrain the representation of aerosol-cloud interactions and their impact on climate forcing.

I also co-coordinate an international, multi-model perturbed-parameter ensemble inter-comparison (PPE-ACI MIP) spanning ECHAM6-HAM, ICON-HAM, UKESM1/HadGEM, CAM6, and other climate models. My responsibilities include developing and maintaining automated Python and Bash pipelines for processing more than 100 TB of climate model and satellite NetCDF data, with version-controlled workflows using Git. I also mentor students in climate model development, scientific computing, and data analysis at SRON, alongside international collaborators.

By combining satellite observations, climate modelling, and advanced machine learning, my research aims to reduce uncertainties in climate projections and improve our understanding of aerosol and cloud processes and their impacts on the Earth's climate system.